StrangerDanger?RemovingBarrierstoSocialConnectionthroughMeaningfulConversation

MarkFishmanAdvisor:WendiGardner

SeniorThesisforMathematicalMethodsintheSocialSciences

NorthwesternUniversityWeinbergCollegeofArtsandSciences

2

AcknowledgementsThis paper would not be possible without the support and guidance of several

people. First, thank you to Professor Joseph Ferrie and Nicole Schneider of the

MMSSProgram,forcoordinatingthethesisprogramandassistingmyselfandpeers

over the past year. To Angela Hosbein, thank you for helping develop Project

Conversation and bringing these ideas to life.Most importantly, thank you tomy

advisor, Professor Wendi Gardner, who met my ambitious ideals and

procrastinationwith trust, patience, and generosity. Finally, tomy parents, thank

youforgrantingmetheopportunityand independenceto followmycuriosityand

pursuemypassionsincollegeandbeyond.

3

Despite our social nature, people often avoid interactingwith strangers in public.

Even when conversation ensues, it usually remains in the realm of “small talk”,

althoughengaginginmoremeaningfultopicsmayhavelastingeffectsonwell-being.

This paper explores beliefs about engaging in meaningful conversation with

strangers, in addition to comparing these beliefs with actual experience.We find

that people choose to avoid social interaction, and when forced to interact,

overwhelmingly prefer to discuss shallow topics rather than deeper material. In

contrasttothesebeliefs,ourstudiesrevealvalueinhavingdeepconversationwith

strangers,particularlyininstillingmeaningfulnessandfeelingsofsocialconnection.

Wethendeviseanexploratorystudytofacilitatemeaningfulconversationsbetween

strangers, the significant results of which spur a discussion of methodological

considerationsandfutureresearchdirections.

4

StrangerDanger?RemovingBarrierstoSocialConnectionthrough

MeaningfulConversation

If asked to describe human nature in one word, “social” quickly comes to

mind.Theabilitytoformcomplexnetworksandrelationshipsbetweenpeopleisa

defining feature of our species, and it is no surprise that social connections are a

largefactorinourindividualwell-being.Breadthanddepthofconnectionsarenot

onlypredictiveofholisticandsubjectivemeasuresofwell-being,butarealsorelated

to physical health (Eid & Larsen, 2008; Lucas & Dyrenforth, 2006; Cohen, 2004;

Helliwell&Putnam,2004).

Oneinterestingquestionabouthumanbehavior,then,isifsocialconnectivity

is so important, why don’t we interact with others more often in public? In the

smartphoneera, inparticular, it is far fromuncommontoseestrangerswaiting in

lineor sittingon the subwaysideby sidewithouta singleverbal exchange.Epley

and Schroeder (2014) attempted to answer this question by exploring the

calibration between beliefs and experiences regarding interacting with strangers.

Their findingssuggest thatdespitewantingtoconverse themselves,peopleexpect

that nearby strangers are less inclined to interact. This “pluralistic ignorance”

dissuades interaction (Miller & McFarland, 1991; Vorauer & Ratner, 1996), even

thoughsuchexperiencesarebeneficialtobothparties.

While Epley & Schroeder (2014) revealed the value of interaction with

strangers in shared environments, they did not control or record the content of

participants’ conversations. It is possible that the type of conversationmoderates

5

the successof an interaction, and if socialnormsdictate theway inwhichpeople

interact (or avoid doing so), then they likely influence the topics of these

conversations.Anafterthoughtsurveyofparticipantssuggestedthatmanyengaged

in what could be considered “small talk”, which aligns with traditional

understanding of social interaction. Asking a perfect stranger about a deeply

personalmattercanbeconsidereduncouth,andforthesakeofsocialharmony,itis

best to stick to “safe” topics. Small talk has an important social role to “define a

mutuallynon-threateningrelationship”andthusisacommonoccurrencebetween

twopeoplewithnosharedknowledgeofeachotherbesidetheirsharedpresencein

timeandspace(Coupland,2003).

Inmaintainingsocialorder, small talk isan important tool,butengaging in

lighter conversation is not an end goal in itself. Spendingmore time engaging in

meaningful dialogue – reflecting on themes beyond everyday episodes – is

correlated with happiness. As a percentage of conversation, the happiest people

spendone-thirdtheamountoftimeengaginginsmalltalk,andtwiceasmuchtime

in meaningful conversation, as the unhappiest among us (Mehl & Vazire, 2010).

Given the difficulty of recording and categorizing conversation over extended

periods, sparse research has been able to test a causal relationship between

meaningfulconversationandwell-being.

Thispaperintroducestheideaofengaginginmeaningfulconversationwith

strangers. We examine existing beliefs about social interaction and meaningful

conversation (Study 1), as well as comparing these beliefs to actual experience

(Study 2). In addition to discussing the theoretical costs and benefits of these

6

experiences,we develop a uniquemethodology to facilitatemeaningful in-person

conversationwithstrangers(Study3).

Study1

The objective of Study 1 was to establish the innate preferences of

participants regarding social interaction with others. First, echoing Epley &

Schroeder(2014),wesoughttodetermineifpeopleinsharedcircumstancessought

isolationorinteraction.Assumingtheywouldchoosetheformer,wethenchoseto

examine the extent to which these preferences were shaped by the nature of

potentialsocialinteraction.

Specifically, given the influence of social norms, we honed in on attitudes

towarddifferenttypesofconversationtopics.Assmalltalkistraditionallyemployed

to assist social interaction between relative strangers, we sought to identify the

presence of a bias towards shallower topics. If forced to interact with another,

wouldparticipantsbemorewillingtoengagein“smalltalk”,suchasdiscussingthe

weather or music preferences, than they would be to choose deeper and more

emotional topics? In addition, we hoped to identify which topics were most and

leastfavored.

We hypothesized that given the choice, participants would avoid social

interaction. Furthermore, if conversation with another were required, preference

would be expressed for avoidingmore emotional and difficult themes in favor of

shallowertopics.

7

Methods

Participants

Seventy-one participants were recruited online using Amazon’s Mechanical Turk

(MTurk) platform, an onlinemarketplace that provides access to a large, diverse

population of workers that can complete a variety of short tasks at their own

convenience. Potential participantswere told theywould be completing a survey

abouttheirpreferencesregardingMTurkstudies.

Of the 71 that completed the survey, all but one answered each of the

questions.Thisperson’sdatawasmostlyincomplete,andthuswasdiscardedforthe

purposeof analysis.Of the remaining70participants, 41weremale and29were

female.

Procedure

Participantswereaskedaseriesofquestionsabouttheirpreferencesrelated

topotentialMTurkstudies.First,eachpersonindicatedwhethertheywouldprefer

to participate in an MTurk study alone or with another person. Afterwards,

participants were presented with eight topics, four of which were considered

“shallow”(e.g.movie/TVpreferences,weather),andfourofwhichwere“deep”(e.g.

dreams for the future, issues needing emotional support). For each topic,

participantsratedthelikelihood(1-7)withwhichtheywouldchoosethetopicfora

hypotheticalconversationwithanotherMTurker.Finally,participantswereaskedto

ranktheeighttopicsintheorderinwhichtheywouldprefertodiscussthemwith

anotherMTurker,rankingthemostdesirabletopicfirstandtheleastpreferredlast.

8

Results

Participants expressed an overwhelming preference for participating in

MTurk studies alone, versus interacting with another person. A full 60 of the 70

participantspreferredisolationinthiscontext.Thiseffectdidnotdifferwithgender,

as85%ofmenand86%ofwomenchosesolitude.

When asked about the likelihood that they would choose various

conversation topics,participantsexpressedapreference for theshalloweroptions

(Figure1).Ratings foreachof the individual topicsarepresented inFigure2.The

effect of conversation typewas significant, F(1,68) = 53.94, p = .000, and did not

differwithgender.

1

2

3

4

5

6

7

Shallow Deep

Figure1:LikelihoodRatingsbyConversationType

Male

Female

Overall

9

The ranked order of conversation topics reinforced these results.

Participantsagainexpressedapreferenceforshallowertopics,F(1,68)=119.73,p=

.000, and this trenddidnotdifferbygender (Figure3).Theaverage rankof each

topicispresentedinFigure4.

0

1

2

3

4

5

6

1 2 3 4 5 6 7 8

Figure2:LikelihoodRatingsbyConversationTopic

Shallow

Deep

1

2

3

4

5

6

7

8

Shallow Deep

Figure3:AverageRelativeRankingbyConversationType

Male

Female

Overall

Note:1ismostpreferable,8isleastpreferable

10

Discussion

Whiletherewasaclearpreferenceforsolitudeoverinteraction,thiseffectis

likely partially driven by the context underpinning the choice. Participants were

askeda general question about theirparticipation inMTurk studies, and thus the

choiceofsolitudelikelyencompassedfactorssuchasefficiencyofcompletingtasks.

TheMTurkplatformprovides individualswith anopportunity to completemicro-

tasksforfinancialcompensation,andparticipantsmayhavebelievedthatworking

withanotherpersonona taskwouldtakemoretimethancompleting italone. (In

Epley & Schroeder (2014), participants believed that talking to a strangerwould

result in a less productive commute, although those that were assigned to do so

rated their commutes as more productive.) Of course, tasks that are longer and

require authentic participation (i.e. collaboration) usually incentivize accordingly.

Giventheprevious literature,andthestaggeringrateatwhichparticipantssought

1

2

3

4

5

6

7

8

1 2 3 4 5 6 7 8

Figure4:AverageRelativeRankingbyConversationTopic

Shallow

Deep

Note:1ismostpreferable,8isleastpreferable

11

to avoid interaction,webelieve other factors besides economic efficiencywere at

playinthisbias.

Thepreferenceforshallowertopicsofconversationwassignificantfromthe

likelihoodratingsalone,buthavingparticipantsrankthetopicsinorderprovideda

clearerpictureof this trend.Figure4clearlydemonstrates thebias,with the95%

confidenceintervalsofeachtopicintheshallowcategoryfullydistinctfromthoseof

the deeper variety. Collecting a relative ranking of the conversation topics also

allowedustoidentifythemostandleastfavorabletopics,whichweusedinStudy2.

Emotionalsupportwasdefinitivelytheleastpreferredconversationtopic,whilewe

chosefoodpreferencestorepresent“smalltalk”,asitwasrankedfavorablybyboth

genders.

Study2

After establishing a significant preference for isolation – or if unavoidable,

shallow conversationwith another –wewanted to compare these beliefs against

actual experiences. Using the first set of responses,we set up an opportunity for

new participants to have an experience either alone orwith a stranger, engaging

witheitherashallowordeeptopic.

JustasEpley&Schroeder(2014)showedthatengagingwithstrangerswas

enjoyabledespiteabiasforsolitude,wehopedtofindthatmeaningfulconversation

was similarly worthwhile, contrary to clear preference for lighter topics. While

thesebiases–forsolitudeandforshallowconversation–aresignificant,wehoped

thatchallengingbothwouldbebeneficial.Despitethefearofdiscussingpotentially

12

uncomfortabletopics,havingdeepconversation,webelieved,mayinstillasenseof

meaninginparticipants,aswellasformabondbetweenconversationpartners.

OurhypothesiswasthatcontrarytothepreferencesestablishedinStudy1,

participantswould actually experience talking to a stranger andengaging indeep

conversation topics more beneficial than experiencing solitude or shallow

conversation.Weexpectedthatindependentoftheeffectsofsocialinteraction,deep

topicswouldincreasemeaningfulnessandfeelingsofsocialconnection.

Methods

Participants

Werecruited212newparticipantsonMTurkforthisroundofthestudy.At

the endof the study, participantswere asked if theyhad completed the activities

seriously, allowing them to withdraw their data from the experiment with no

financial repercussion. From this question, we removed 14 participants from

analysis,leaving198validresponses.

Procedure

Participants were assigned to one of four conditions according to a 2x2

between-subjects design. Each participantwas assigned to experience solitude or

interaction, inadditiontodeepvs.shallowtopics.BasedontheresultsofStudy1,

we selected one of the least preferred deep topics – emotional support – as the

“deep”topic,andoneofthemostpreferredshallowtopics– foodpreferences–as

the“shallow”topic.

13

In the social interaction assignments, participantswere asked to read and

respondtooneoftwoshortlettersfromafellowMTurker(althoughwrittenbythe

researchers).Forthosealsoassignedtothedeeptopic,thisletterdiscusseddealing

withanemotionalstruggle,whereasfortheshallowtopic,thelettermentionedfood

preferences. For context, below is the letter shown to participants in the

intersectinginteractionanddeeptopicassignments.

“Hi,

MynameisK,andI’mfromSanFrancisco,currentlyinthemidwest.I’m34and

Iamaneconomicsspecialist.I’mfairlyactive–Iliketobike,swim,anddoyoga.

Lately, I haven’t had the energy to be active though... You seemydogMabel

passedawayrecentlyandI’vebeenfeelingreallydepressedeversince. Idon’t

even feel likemyselfanymore.All Ido ismopearoundallday. Ibarelyeator

sleep.IgotMabelwhenIfirstmovedoutinmyearlytwenties.It’sbeenreally

hardadjustingtolifewithouther.Thehouseseemssoemptynow.Imisshera

lot. Do you have any pets? Anything you need emotional support for? I'd be

happytolisten.

Hopetohearfromyousoon,

K”

Whilenotspecificallyinstructedtodiscusstheirownemotionalstrugglesor

foodpreferences,weaskedparticipantsinthesocialinteractionconditionstoreply

totheletterswithafewsentences,hopingtheywouldnaturallyrespondontopic.

14

Thoseinthesolitudeconditionsdidnotreceivealetterfromafictitiouspen

pal,butwerealsoasked towritea shortnoteon the topic corresponding to their

deepvs.shallowcondition,imagininganotherMTurkerwouldreply.

Afterwritingthenote,allparticipantswereaskedaseriesofquestionsabout

theexperienceand their current feelings.Eachparticipantwasasked to ratehow

enjoyable,meaningful,andengagingthetaskwas,aswellashowsociallyconnected

itmadethemfeel.UsingthesamemeasuresasEpley&Schroeder(2014),wealso

askedthemtoprovideratingsofhowhappy,sad,sociallyconnected,andlonelythey

presentlyfelt.Alloftheseratingswerepresentedona1-5scale.

Results

Figures5-8depicttheeffectsofeachconditiononthefourmaindependent

variables. Participants in the social interaction condition had more positive

experiences than those who were not responding directly to a letter. These

differences were significant in terms of enjoyment F(1,194) = 18.87, p = .000,

meaningfulness,F(1,194)=9.35,p=.003,andfeelingsofsocialconnection,F(1,194)

=11.32,p=.001.Althoughthedifferencewasnotsignificant,thedirectionofmeans

suggests that discussing foodpreferenceswasmore enjoyable thanwriting about

issues needing emotional support. Participants found writing about emotional

issuestobemoremeaningful,F(1,194)=9.01,p=.003.Levelsofengagementwith

thetaskdidnotvaryacrosstheconditions.

15

1

2

3

4

5

Reply Alone

Figure5:Enjoyment

Deep

Shallow

1

2

3

4

5

Reply Alone

Figure6:Meaningfulness

Deep

Shallow

1

2

3

4

5

Reply Alone

Figure7:Engagement

Deep

Shallow

16

Tofurtheranalyzetheresults,participantswhowereinboththeinteraction

and deep topic conditions were compared to those in the other three conditions

using3v1weightedcontrasteffects.Thecombinatoryeffectoftheseconditionswas

predictive of both meaningfulness, t(194) = 3.91, p = .000, and feelings of social

connection,t(194),=2.10,p=.037,andmarginallysignificantforenjoyment,t(194)

=1.83,p=.068.

Following the model of Epley & Schroeder (2014), we combined the two

moodquestionsintoanindexofpositivemoodbyreversescoringthesadnessscore

andaveragingitwiththeirhappinessscore.Wealsocombinedthetwoconnection

itemsinasimilarwaybyreversescoringlonelinessandaveragingitwithfeelingsof

connection. To analyze participants’ feelings, a mixed model repeated measures

ANOVA was used with type of feelings (mood or connection) as within-subject

factors, and the social condition (social interaction vs.writing a letter alone) and

1

2

3

4

5

Reply Alone

Figure8:SocialConnection

Deep

Shallow

17

letter topic (shallow vs. deep) as between-subject factors. Analyses revealed a

significantinteractionoftypeoffeelingbysocialcondition,F(1,194)=7.38,p=.007,

andbetween typeof feelingsand letter topic,F(1,194)=7.74,p= .006.Ascanbe

seeninFigure9,participantsfeltmorepositiveafterwritingorreplyingaboutfood

preferencesthanforanissueofemotionalsupport(likelyduetothenegativenature

oftheeventrequiringsupport).Figure10displaystheeffectofeachmanipulation

onsocialconnection,forwhichonlysocialconditionwassignificant.

1

2

3

4

5

Reply Alone

Figure9:Mood

Deep

Shallow

1

2

3

4

5

Reply Alone

Figure10:Connection

Deep

Shallow

18

Discussion

The results of Study 2 confirmed the benefits of social interaction, and

revealed the nuances of grappling with difficult material. We believe that

engagement with the task was not a precise enough measure for this study,

consideringthatmostMTurktasksarefarmorerepetitiveand/orimpersonalthan

any of the writing exercises completed by our participants. All participants were

askedtowriteafewsentencesaboutthemselves,ataskthat,whencomparedwith

other MTurk studies, understandably received relatively high marks for

engagement.Apartfromengagement,theeffectofsocialinteractionwassignificant

acrosstheboard.

Comparing the responses for deep and shallow topics was slightly more

difficult, as the effects of this manipulation were only significant for ratings of

meaningfulness. Isolating the effect of isolation vs. interaction,we did not expect

anydifferencesinfeelingsofsocialconnectionbetweenthetwotopics.Inaddition,

whilediscussingemotionallydifficultsituationsmaybeameaningfulexperience,it

is not necessarily an enjoyable one. The promise of small talk (in general) is

engaginginnon-offensiveandeasilydigestibleconversation,andintheshort-term,

itisnotsurprisingthatpeoplemayprefertotalkabouttheirfavoritefoodthanthe

rockystateoftheirmarriage.

After teasing out the individual effects of these conditions, we wanted to

examineourresearchquestioninbroadterms–isengagingwithastrangerindeep

conversation beneficial? Contrast effects revealed that such an experience,

compared to those of the other participants, inspired both meaningfulness and

19

feelings of social connection, in addition to marginally significant enjoyment. If

meaningfulnesscanbethoughtofasatypeoflong-termbenefit,whileenjoymentis

moreofanimmediateboost,thissuggeststhattheenjoymentofinteractingwitha

strangermayovercomethedifficultiesofengagingwithdeepermaterial.Itmaybe

that approaching a stranger is somewhat easierwhen one can hover around safe

andmutuallynon-intrusivetopics.Yetdiggingabitdeeperintoaconversationwith

astranger,webelieve,maybelessriskythanwithafriend:theknowledgethatone

mayneverseetheotherpersonagainmayprovideanopportunityforanhonestand

vulnerableexchange.

Themaindifficulty for theexperimentaldesignwas findinganappropriate

controlfortheisolationcondition.Whatisthereal-lifealternativetoengagingwith

a stranger in sharedcircumstances?Onemayscroll throughsocialmediaon their

phone, read a book, listen to a podcast, or call a friend or familymember on the

phone.Ofcourse,onecanalsorejectallextraneousstimuliandremainsilent,paying

attention to their environment or entertaining whatever thoughts pass through

theirmind.Whataretherelativecostsandbenefitsoftheseactivities,andhowdo

theycomparetoengaginginconversationwithastranger?Thisdifficultyofdealing

with this ambiguity was compounded by the fact that our participants were

restricted to online tasks. To isolate the effect of social interaction, we had

participantsintheisolationconditionthinkandwriteaboutthesametopicsastheir

counterparts,withtheexceptionthattheywerenotdirectlyrespondingtoanother

person’sthoughtsandexperienceswiththesubject.Thiswouldbecomparable,we

think, to someone journaling on the subway, an activity that is likely at least as

20

enjoyable,meaningful,andengagingthanwhatmanycommutersmindlesslyengage

in. While of course there are plenty of meaningful activities in which captive

individuals can participate, one likely only has to look at those in line at a coffee

shoptobelieveourchoiceofcontrolconditionprovidesaconservativeestimateof

therelativebenefitsofengaginginmeaningfulconversationwithstrangers.

GeneralDiscussionofStudies1-2

The primary motivation for conducting Studies 1 & 2 online was to

accumulatea largeenoughsamplesize to illuminatepotentiallysignificanteffects.

Study 2 required enough participants for a between-subjects design with four

conditions,andwealsodeemeditnecessarytorecruitdifferentpopulationsforthe

twostudies,astoavoidanywithin-subjectconfounds.(Theorderedpreferencesof

conversationtopicswerealsousedtodesignthelettersinStudy2.)Theresultsfrom

Study1wererobustenoughforustoassumethepreferenceswererepresentative

ofageneralpopulation,allowingustodirectlycomparethebeliefsofthefirstsetof

participantstotheexperiencesofthesecondgroup.

Themajorlimitationofonlinestudies,ofcourse,isthattheinteractionsare

not representativeof thesocial interactionsof interest for theresearchquestions.

(Also noteworthy is the fact that, since we would be asking about participants’

experiences on MTurk in Study 2, we asked about participants’ preferences

specifically related to MTurk in Study 1, a decision that prioritized internal

consistency rather than external validity. In reality, we were not interested in

people’s socialpreferences aboutMTurk,butwanted tohaveadirect comparison

21

between preferences and experiences between the two studies.) For the sake of

control,welimitedcommunicationtoonenoteofafewsentences,withparticipants

receiving no feedback to their letter. Unlike real conversations, there was no

continuousback-and-forth, and requesting a truncated cross-sectionof a dialogue

wasunlikelytoinspirethesamedepthofcommunicationasaconversationinwhich

questions could be both asked and answered. While many participants took the

writingactivityseriously,mostof themlikelyknewthat theirnotewouldproduce

noresponse,andpossiblythatthelettertheyreceived(forthoseintheinteraction

conditions) was fabricated. Further missing were all of the visual, auditory, and

other non-verbal cues that animate an offline social interaction. Much is

communicated through eye contact, body language, and vocal inflection, and we

expect that conducting a similar, yet offline, experiment – with longer and more

genuineinteractions–wouldproducemoresignificantandenlighteningresults.

Although other methodological designs were considered, such as having

multiple MTurkers communicate in real-time or chat with a confederate, these

would have presented their own set of challenges. Studies 1& 2were ultimately

effectiveinhighlightingandsupportingourhypothesesaboutthemisguidedbiases

for isolation and small talk in the presence of strangers. Yet the limitations

mentioned above were motivation for a different approach to our research

questions.

22

Study3

While Studies 1 and 2 provided some theoretical evidence for our

hypotheses, we sought to further investigate our research question with a more

externally valid approach. Referring back to our original question of why people

don’tinteractinpublic,wewantedtomovebeyonddigitalplatformstostudyactual

in-person interactions.This isparticularly important inaworld thatenablesus to

communicate through an increasing number of digital mediums, often without

seeinganother’sfaceorhearingtheirvoice–ifoneiscommunicatingwithahuman

at all. Trading control for validity, we sought to create an exploratory study that

deviated from themethodology of the previous experiments. As discussed in the

previous section,measuring the relative benefits ofmeaningful conversationwith

strangersisdifficultduetotheambiguousnatureofacontrolcondition.Insteadof

comparing effects across conditions,wewanted to explore if the experience itself

wasperceivedasworthwhiletoindividuals.Theprimaryquestion,then,forStudy3

washowtoputourtheoretical findingstowork.Knowingdeepconversationwith

strangerswasameaningfulexperience,howcouldweovercometheinherentbiases

forsmalltalkandsolitude?Isthereaneffectivewaytoencourageandfacilitatesuch

experiencesoutsideofanexperimentaloracademicsetting?

These questions prompted a collaboration with “Project Conversation,” a

student organization whose mission was to connect undergraduate students and

facilitategenuineconversationsoncampus.WeassistedProjectConversation(P.C.)

as the organization iterated and refined itsmethodology for the better part of an

academic year, before officially researching participants. In striving to facilitate

23

social interaction, the foundational design premise of P.C. became overcoming

pluralistic ignorance. According to Epley & Schroeder (2014), who identified this

phenomenon as amajor reasonpeople avoid public social interaction, “pluralistic

ignorance, whereby people consistently think others are less interested in

connecting than they are themselves, not only could make an attempted

conversationseemunpleasantbutcouldalsocreateabarriertolearningthatone’s

expectations aremistaken” (pg. 1986).Tokeep suchbeliefs from inhibiting social

interaction, we designed an opt-in system that removed the burden of initiating

interaction fromthe individual.Bysigningup,studentswereguaranteedtomatch

withsomeonewhohadalsoexpressedadesiretoparticipate.Weexpectedthatsuch

asystemwouldremovethepotentialawkwardnessofapproachingastrangerand

engaginginconversation–adiscomfortusuallyaddressedbyparticipatinginsmall

talk.Thisdesignshouldthusnotonlyhelpinitiateconversations,butalsoallowthe

participatingindividualstoopenthemselvestomoredifficultandintimatetopics.

Methods

Participants

Studentswere recruited for the studymainly throughword ofmouth and

onlinesocialmedia.InformationaboutP.C.wasdistributedthroughFacebook,and

experimenters also notified classmates about the opportunity. P.C. wasmarketed

not as an academic study, but as a student-led project that facilitatedmeaningful

conversations between strangers. The sign-up form contained the following

introduction,withfurtherdetailedinstructionsonhowtoparticipate:

24

“We started Project Conversation to give students an opportunity to have

meaningful conversations with students across all of campus. We provide a

platformforyoutotalktosomebodynewwithouttheawkwardnessandsocial

stigmaof reachingout to strangers. It'snotadatingappora friendmaking

service-it'sabouttheconversation,andit'snotexpectedthatyoustayintouch

afterwards.”

In the initial fewmonthsofmethodological tests,31studentspartook inat

leastoneiterationofP.C.Whenformalresearchcommencedinthespring,60sign-

ups were garnered over a three-week period. Due to individual logistical and/or

communicationissues,only42continuedtointeractwithanotherstudent,with38

ofthesecompletingthepost-interactionfeedbackformneededfordataanalysis.Six

studentsparticipatedtwice,andtheirsecondroundofdatawasexcluded.Thisleft

32uniqueparticipantswhocompletedthepost-conversationsurvey.

Procedure

Participants signedup througha link thatdirected them toanonline formwitha

short participation survey.University emailswere required to viewand complete

theform,butnameswerenotrecordedwithsubmission.Theformcollectedcontact

information(emailandphonenumber);basicdemographicfactorssuchasgender,

race/ethnicity, andyear in school; answers to a fewmoodquestions; and specific

availabilitytoparticipateintheexperiment.AdaptingmeasuresagainfromEpley&

Schroeder (2014), we collected pre-test measures of happiness, sadness, and

feelings of social connection. Participationwas limited to five one-hour timeslots

that were pre-selected for the week by the experimenters. These timeslots were

25

randomlydistributedacrosstheweek(oneoneachweekday)duringtimesavailable

totheexperimenters.Respondentsindicatedavailabilityforeachofthesetimeslots,

and were notified via email by Sunday night of the timeslot to which they were

assigned for theupcomingweek. (Ifparticipantswereunable tobematched, they

wereasked toparticipateduring thenext roundof conversations.)The time slots

wereupdatedeachSundayforthefollowingweek.

Eachmorning,anautomatedtextmessagewassenttoparticipantsscheduled

tomeet thatday,using the serviceTextMagic.Themessagecontained the specific

locationformeetingtheirpartnerattheassignedtime(alllocationswerewithinthe

university student center). Participants were also instructed to respond to the

message when they had found their partner, which prompted an automatic

response from TextMagic that contained a link to further instructions.

Experimenters used TextMagic to communicate in real-time with participants if

needed,forexample,ifaparticipantwasrunninglate,orinextremecases,neededto

reschedule.

When matched pairs identified each other at their assigned meeting spot,

they found a place to sit for the next hour, and replied to the morning’s text to

receive further instructions.Theensuingauto-response text linked toadocument

with a general description of the next hour’s activities. First, the document asked

participants to avoid conversation that they would consider “small talk”, and

suggestedthatconversationdoesnotrevolvearoundstudentlifeattheuniversity,

such as classes and extracurricular activities. To inspire deeper conversation, the

document featureda fewquestions thatparticipants could choose todiscusswith

26

eachother.Thedocumentwasupdatedeachweekwithdifferentthemedquestions,

coveringtopicssuchasfamilyrelationships,hopesandfearsforthefuture,andhow

onehaschangedsincestartingcollege.Itwasnotrequiredthatparticipantsanswer

these exact questions, although participants could use them as they wished. In

addition,thedocumentremindedparticipantsthattheirinteractionwasscheduled

for one hour, and that they would receive a final automated text message five

minutes before the hour was up. This final message included a link to the post-

interaction survey, which participants were asked to complete immediately.

Responses were also required for entrance into a drawing for one of several

Amazongiftcards.

This final survey collected participants’ thoughts about the interaction, as

wellas themeasuresofhappiness,sadness,andfeelingsofsocialconnection from

before.Whilethesurveywasanonymous,participantsenteredtheirphonenumber

on this survey in order to match responses to pre-test measures and confirm

participation in thedrawing for compensation.Participantsweregivenno further

instructions about their interaction, and, after completing the final survey, were

allowedtocontinuetheconversationwiththeirpartnerifdesired.Weexpectedthat

some partners would exchange contact information and form relationships after

thispoint,butidentifyingfurtheroutcomeswasnotthepurposeofthestudy.

Results

Twosetsofdatawereusedtoanalyzetheexperience.First,within-subjects

paired t-testswere used to compare participants’ change inmood and feelings of

27

socialconnection,ascollectedduringsign-upsanddirectlyaftersocial interaction.

Additionally, participants were asked about their holistic experiences about the

project.Thepost-interaction formcollectedmeasuresofenjoyment, the likelihood

ofparticipating inasimilarexperienceagain,and the likelihoodofrecommending

suchanexperiencetoafriend(1-7).

Feelings of happiness and sadness were recoded into a composite mood

measure. The changes in mood and feelings of social connection are depicted in

Figure 11. The effects of conversation on bothmeasureswere significant, t(31) =

3.40,p=.002formood,t(31)=2.64,p=.013forsocialconnection.

Holistically, the large majority participants had favorable experiences.

ResponsesforholisticmeasuresareshowninFigures12-14.Themedianresponse

foreachofthethreequestionswas7,withonlyahandfulofparticipantsreporting

lessthanidealexperiences.Additionally,allbutoneoftheparticipantsaskedtobe

1

2

3

4

5

6

7

Mood SocialConnection

Figure11:ChangesinMoodandSocialConnection

Before

After

28

notified if/when future rounds of P.C.were open for sign-ups. (The single person

whopassedindicatedthatshewouldbegraduatinginafewweeksandlikelywould

notbeavailable.)

0

5

10

15

20

1 2 3 4 5 6 7

Figure12:EnjoymentofExperience

0

5

10

15

20

1 2 3 4 5 6 7

Figure13:LikelihoodofRepeatingExperience

29

Discussion

As measured by the increase in mood and feelings of social connection,

participants found theexperience tobebeneficial in themoment.Whilemeasures

werenotrecordedata later time, results fromMehl&Vazire (2010)suggest that

increasingtheamountofmeaningfulconversationmayboostindividualwell-being

over a longer time period. Although most participants expressed a desire to

participate in a similar experience again, there is evidence that may suggest one

conversation is enough to sustain long-term benefits. Participants said theywere

slightly more likely to recommend the experience to a friend than to participate

again, and thedifferencebetween thesemeanswasmarginally significant, t(31)=

1.98, p = .056. If one understands the likelihood of recommendation to be an

objective measure of value (assuming participants have friends’ best interests at

heart),thentheslightdecreaseinlikelihoodofpersonallyrepeatingtheexperience

representsadecreaseinmarginalvaluefortheparticipant,suggestingthatthefirst

0

5

10

15

20

25

1 2 3 4 5 6 7

Figure14:LikelihoodofRecommendingExperience

30

experience may be more important than subsequent conversations. Similarly,

participants provided higher levels of recommendation than levels of enjoyment,

t(31) = 1.82, p = .078, suggesting some recognition of the objective value of

participating in the experience, despite one’s personal experience. Overall, more

than90%ofparticipantsreporteda6or7(ona7pointscale)forthelikelihoodof

recommending the experience to a friend (with the other three participants

providinga5),suggestingpeopleunderstandthevalueofengaginginameaningful

conversationwithastranger.

While certainly encouraging, there are a few reasons for a bit of healthy

skepticism about the significance of the results. First, while large enough for

statistical significance, thesamplesize isnot robust.Fourpeopleparticipatedand

did not fill out the post-interaction survey, potentially biasing the results.

(Anecdotally, though, we know that these participants had very positive

experiences:whenaskedlater,bothpairsreportedstrikingupalongconversation

withouteverlookingattheinstructions.)Thehighattritionrateamongparticipants

– many had last-second emergencies, also voiding the experience for would-be

conversationpartners–isalsoapotentialcauseforconcern,yetwehavenoreason

tobelievethatthosewhonevermettheirpartnerswouldbemoreorlesslikelyto

benefitfromtheexperience.(Manyofthesepeopleendedupparticipatingthenext

weekwithanewpartneranyway.)

Even if Study 3 consisted of a larger sample, the population from which

participants were recruited still serves as a limitation. All participants were

undergraduatestudentsat thesameuniversity,andthuswereguaranteedtohave

31

muchincommonwiththeirconversationpartners.Althoughresultsfromthepost-

interactionsurveyconfirmedthatallpairswereperfectstrangers,eachparticipant

entered the interaction with a shared set of experiences and institutional

vocabulary.Thisintheorywouldmakesmalltalkquiteeasy–participantscouldask

about classes or extracurricular activities – but instructions encouraged them to

engage in more meaningful conversation. Still, it is likely that there was an

understoodcommonalitybetweenpairs,providingafoundationoftrustandsocial

connection that could facilitate more enjoyable and intimate conversation. This

guaranteeofmeetingwithafellowundergraduatestudentmayhavealsoinfluenced

theperceivedriskof the interaction.Ononehand, students faced lessuncertainty

about theperson theywouldmeetwith (compared toapartnerselected fromthe

population at large). The flipside of selecting from a contained population is that

whilepartnerswerelesslikelytobecompletestrangersbeforemeeting,therewas

also a chance of encountering each other in the future. Revealing secrets to a

strangeron thesubway inNewYorkCitypromisesa levelofanonymity that isn’t

guaranteed in talking to someone who may be in a future class of yours or

eventually date a mutual friend. Finally, participants were volunteers who

expressedinterestinhavingameaningfulconversationwithastranger,soitcannot

be assumed that any student chosen at random would have as positive of an

experience as those reported by participants. Unlike Studies 1-2,which recruited

randomanddiversesamples,Study3cannotpromiseamoreuniversalenjoyment

ofthetreatment.

32

Project Conversation and themethodology of Study 3were designed for a

specificpopulationandenvironment.Thus,asdiscussedabove, theresultsarenot

entirely representative of how strangers in other contexts would experience

meaningfulconversationswitheachother. ImprovementstoP.C.couldbemadeto

betterfitalargerandmorediversepopulation(e.g.developingamobileplatformto

streamline matching and communication), but in some ways the study is

fundamentallydifferentthanhowstrangersnormally interact.Onemajorquestion

istheissueoftiming.ParticipantsofStudy3wererequiredtoblockanhouroftime

for a conversation, whereas natural conversations with strangers are usually

spontaneousandfleeting.Thisisprobablypartiallyduetotheshallownatureofthe

conversation (how long do youwant to talk about theweather?), but the shared

circumstancesthatpromotetheseinteractionsarealsousuallynotsharedforlong

(how likely is someone to share their childhood dreams before they order their

coffee?).Unlesstwopeoplehavetimetospareandcaninhabitthesamespaceforan

extendedperiod (maybe ina caféor abookstore), it isunlikely that ameaningful

conversation will flourish spontaneously. In a “real-world” interaction, one also

usuallyhas theability toopt-outof theconversation–althoughsocialnormsmay

makethishard,say,onanairplane–whileourparticipantswerecommittedtoat

leastthehourtheyhadsignedupfor.Ifonebelievestheonlywaytounderstandthe

benefits of meaningfully interacting with strangers is to perfectly replicate the

mannerinwhichmost interactionsnaturallyoccur,thenStudy3wouldnotbethe

ideal research technique. Yet P.C. provides an alternative to the traditional

approach,creatingvaluethroughthemethodologyitself.Thus,whileasystemlike

33

P.C. is not fully representative of how strangersmay interact in the “real world”

today, it at least presents opportunities for further research, and potentially an

opportunityforhowstrangerscanmeaningfullyinteractinthefuture.

Conclusion

This work provides further evidence that despite our inherently social

nature, people often choose not to interact with others, especially in meaningful

conversation–eventhoughthis isapositiveandvaluableexperience.Engaging in

conversationthatventuresbeyondsmalltalkmaybeintimidating,especiallywitha

completestranger,butitisoftenworththisrisk.

Thispaperalsohighlightsseveralopportunitiesforfutureresearch.Thereis

still a dearth of literature on the immediate and long-term effects of small talk

versusmeaningfulconversation,andthisisoneofthefirstattemptsatinvestigating

thesedynamicswithconversationsbetweenstrangers. Inparticular,relatedtothe

discussion of population samples in Study 3, it would be worth exploring how

beliefsaboutmeaningfulconversationdifferdependingonthe familiaritybetween

conversationpartners.Aconversationwithacompletestrangermayactasahurdle

toestablishingsharedgroundbutalsoasapromiseofanonymity,andoneofthese

factorsmaybestrongerinpromotingauthenticityandvulnerabilityinconversation.

Additionally, larger sample sizes should allow future research to identify trends

relatingtopartnercharacteristics.Forexample,dopeoplebenefitmorefromtalking

tosomeoneofadifferentraceorgender?Figuringoutwhobenefitsmostfromthese

34

conversationscouldopenthedoortocultivatingempathyandunderstandingacross

populations.

AmethodologylikethatemployedinStudy3isnotthetraditionalresearch

approach for exploring the dynamics of naturally occurring interactions between

strangers. But by following the literature, we have devised the framework for a

platform thatwe believe can contribute to society asmuch as it studies it. If the

ultimategoalofacademicresearchistobetterunderstandtheworldaroundusfor

thesakeofimprovingcollectivewell-being,thenitbehoovesusnottojustobserve

realitybutalsotoalteritwhenapplyingfindingscanservethegreatergood.Ideally,

somethinglikeProjectConversationcanserveasasupplementtothewayinwhich

strangers interact.According to Schroeder&Epley (2014), oneof the reasonswe

underestimate the enjoyment received from talking to strangers is because our

beliefs fail to calibrate to our experiences. P.C. may serve as an opportunity to

facilitate conversations in order to continuously update beliefs, giving people the

confidencetoinitiatesocialinteractionswithstrangersontheirown.

Weliveinaworldthatisincreasinglysupportiveofourinnatebiasesagainst

interacting with strangers and engaging meaningfully with others, as technology

gradually removes risk in favor of comfort and immediate gratification. Although

technologypromisesmoreconnectionthanever,itisthosewhoengagewithothers

offlinewho experience the greatest feelings of connection (Sherman et al., 2013).

For thousands of years, our species interacted solely through face-to-face

conversation,andwhiletechnologysurelyhasitsbenefits(suchassupportinglong-

distance relationships), it is important as ever to realize the importance of

35

meaningful, in-person conversation. As Sherry Turkle, the author of Reclaiming

Conversation, said,we are living “moments ofmore, and lives of less.” An endless

stream of curated content ensures boredom never comes calling, so why risk a

conversation with another person in line for coffee? By immediately plugging in

headphonesontheairplane,weavoiddiscomfortbutpassupthechancetochatup

the person across the aisle who may gift us a new perspective on an ailing

friendship. We supposedly live in a hyperconnected age, but we are constantly

passingupopportunitiesforconnectionwithothersandourselves.Talkingtothose

wefeelweareleastconnectedtomaybeastepforward.Smalltalkisanaturaland

important element of society’s social fabric, particularly when connectingmutual

strangers– contextoftendictates thatpeopleadhere to safe topics.Yet if there is

valueinengagingstrangersinmeaningfulconversation,thenitisworthchallenging

thestatusquotofacilitatemoreoftheseexperiences.

36

WorksCitedCoupland, J. (2003).SmallTalk:SocialFunctions.ResearchonLanguageandSocial

Interaction,36(1),1-6.Cohen Sheldon.2004. “Social Relationships and Health.” American Psychologist

59(SpecialIssue):676–84.Eid,M., & Larsen, R. J. (Eds.). (2008).The scienceof subjectivewell-being. Guilford

Press.Epley, N., & Schroeder, J. (2014). Mistakenly seeking solitude.Journal of

ExperimentalPsychology,143(5),1980-1999.Helliwell, J., & Putnam, R. (2004). The Social Context of Well-Being.Philosophical

Transactions:BiologicalSciences,359(1449),1435-1446.Lucas R, Dyrenforth P. Does the existence of social relationships matter for

subjective well-being? In: Vohs KD, Finkel EJ, editors.Self and relationships:Connecting intrapersonal and interpersonal processes.Guilford Press; NewYork:2006.pp.254–273.

Mehl, M. R., Vazire, S., Holleran, S. E., & Clark, C. S. (2010). Eavesdropping onHappiness: Well-being is Related to Having Less Small Talk and MoreSubstantiveConversations.PsychologicalScience,21(4),539–541.

Miller,D.T.,&McFarland,C.(1991).Whensocialcomparisongoesawry:Thecaseofpluralisticignorance.InJ.Suls&T.Wills(Eds.),Socialcomparison:Contemporarytheoryandresearch(pp.287–313).Hills-dale,NJ:Erlbaum.

Sherman,L.E.,Michikyan,M.,&Greenfield,P.M.(2013).Theeffectsoftext,audio,video, and in-person communication on bonding between friends.Cyberpsychology:JournalofPsychosocialResearchonCyberspace,7(2).

Vorauer,J.D.,&Ratner,R.K.(1996).Who’sgoingtomakethefirstmove?Pluralisticignorance as an impediment to relationship formation. Journal of Social andPersonalRelationships,13,483–506.